Kinetic and mechanistic study of the H-transfer reduction of dimethyl itaconate by a Rh/TPPTS catalyst under biphasic conditions: Evidence for a rhodametallacycle intermediate

Article abstract of DOI:10.1002/1099-0682(200007)2000:7<1495::AID-EJIC1495>3.0.CO;2-J

The H-transfer reduction of dimethyl itaconate under biphasic catalysis using sodium formate/water as the hydrogen source with a Rh/TPPTS complex was studied. Labelling experiments reveal that the reduction proceeds through a 1,3-hydrogen addition. Kinetic studies allow us to propose a catalytic cycle going through a rhodametallacyclobutane intermediate with a rate-determining step involving sodium formate. A formal kinetic rate law was then derived from this mechanism which accounts for the results. This work evidences the switch between the mechanism of diacid reduction and that of the corresponding esters.

Full text of DOI:10.1002/1099-0682(200007)2000:7<1495::AID-EJIC1495>3.0.CO;2-J

FULL PAPER
Kinetic and Mechanistic Study of the H-Transfer Reduction of Dimethyl
Itaconate by a Rh/TPPTS Catalyst under Biphasic Conditions: Evidence for a
Rhodametallacycle Intermediate
Nathalie Tanchoux[a] and Claude de Bellefon*[a]
Keywords: H-transfer / Kinetics / Reaction mechanisms / Biphasic catalysis / Rhodium / Metallacyclobutanes
The H-transfer reduction of dimethyl itaconate under bi-
phasic catalysis using sodium formate/water as the hydrogen
mediate with a rate-determining step involving sodium form-
ate. A formal kinetic rate law was then derived from this
source with a Rh/TPPTS complex was studied. Labelling ex- mechanism which accounts for the results. This work evid-
periments reveal that the reduction proceeds through a 1,3-
ences the switch between the mechanism of diacid reduction
hydrogen addition. Kinetic studies allow us to propose a cata- and that of the corresponding esters.
lytic cycle going through a rhodametallacyclobutane inter-
Introduction
The asymmetric hydrogenation of unsaturated substrates
by either molecular hydrogen or H-transfer reduction is
now one of the most important applications of homogen-
eous catalysis by chiral transition metal complexes.[1–4]
Many studies report the use of optically pure phosphane
complexes of Rh and Ru in organic[1,3] as well as in aque-
ous[2,5] or biphasic media.[6] The mechanisms of these reac-
tions have been extensively studied, but very few details are
known about the H-transfer reduction of α,β-unsaturated
carboxylic esters using sodium formate/water in biphasic
catalysis. As these reactions catalysed by chiral diphosphane
complexes in aqueous medium are generally very complex,
the racemic reduction of a prochiral substrate with a non-
chiral monophosphane ligand was first studied. The kinet-
ics of this reaction using dimethyl itaconate as a substrate
and Rh/TPPTS as a catalytic complex were further investig-
ated,[7] and labelling studies were undertaken in order to
understand the mechanism and propose a catalytic cycle
reflecting both the kinetic and mechanistic studies.
Analysis of the organic layer by gas chromatography and
mass balance calculations confirm the near quantitative
(Ͼ 95%) conversion of the substrate, DMI, into the prod-
uct, DMS. A side product (Ͻ 5%) was detected and identi-
fied as an isomer of DMI, 2-dimethylmesaconate (DMM),
formed by the shift of the double bond from the external
to the internal position. Such an isomerisation of terminal
to internal olefins has been frequently reported as a side
reaction in for example hydrogenations.[8–10]
Liquid–Liquid Partition Equilibrium
As the course of the reaction is more conveniently fol-
lowed by analysis of the organic phase, it is primordial to
correlate the measured concentrations in the organic phase
to the concentrations in the aqueous phase, since the latter
is where the catalytic reaction actually takes place. The con-
centration of species A in the aqueous and organic layers
were described by the linear relationship of Equation 2.
Results
Reaction
The reaction studied is the reduction of dimethyl itacon-
ate (DMI) to dimethylmethylsuccinate (DMS) with sodium
formate, catalysed by a water-soluble Rh complex in a bi-
phasic medium (Equation 1). The catalytic system was gen-
erated by mixing the dimer [Rh(cod)Cl]2 and the sodium
salt of triphenylphosphanetrisulfonate (TPPTS) in an aque-
ous sodium formate solution.
[a]
´
´ ´
Laboratoire de Genie des Procedes Catalytiques,
All the measurements of partition coefficients have al-
ready been described in a previous paper.[7] The coefficients
calculated from the experimental results are given in
Table 1.
CNRS-CPE Lyon, URA 2211,
43, boulevard du 11 novembre 1918, B.P. 2077,
69616 Villeurbanne, France
Fax: (internat.) ϩ33-4/72431673,
E-mail: Claude.de.Bellefon@lgpc.cpe.fr
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502
WILEY-VCH Verlag GmbH, D-69451 Weinheim, 2000
1434Ϫ1948/00/0707Ϫ1495 $ 17.50ϩ.50/0
1495

N. Tanchoux, C. de Bellefon
FULL PAPER
Table 1. Cyclohexane/water partition coefficients of DMI and
As seen in Figure 1, the observed initial reaction rate
shows a first order dependence with respect to the catalyst
concentration for formate concentrations of 5 and 1
kmol m–3.
DMS for various sodium formate concentrations at 293 K; a linear
DMI
partition is observed for 0 Յ C
Յ 0.13 kmol m–3
Org
–3
cHCOONa [kmol m
]
0
1
2.5
5
Aq
–3
PDMI [m3Aq
PDMS [m3Aq
m
]
1.2
1.7
2.8
4
4.3
5.6
11.3
14
Org
–3
m
]
Org
Kinetic Studies in a Well-Mixed Batch Reactor
The absence of mass transfer limitations in a well mixed
batch reactor at 1100 rpm was previously checked.[11] A pre-
vious study[7] allowed the extraction of an activation energy
of 71 kJ mol–1, in good agreement with those reported for
other hydrogen transfer reductions of olefins.[12] The study
also led us to propose the semi-empirical rate model of
Equation 3, taking into account all the results obtained
during that preliminary study.
Figure 1. Influence of the catalyst concentration on the observed
DMI
–3
initial reaction rate riOrg [T ϭ 313 K, C
ϭ 0.155 kmol m
,
Org
Org
TPPTS/Rh ϭ 6; (᭜ and ᭺): α ϭ 1; (∆): α ϭ 0.5; ( ᮀ): α ϭ 0.45)
This model being based on preliminary results, it is only
partial and does not take into account important para-
meters such as the concentrations of sodium formate or
phosphane. More tests varying these parameters were per-
formed in order to propose a more complete kinetic model
that included the new information obtained. The range of
conditions for this study is given in Table 2.
The influence of initial substrate concentration (Figure 2)
on the observed initial rate shows an apparent zero order
for the experiments carried out at a formate concentration
of 1 kmol m–3, whereas it is complex for a concentration
of 2.5 kmol m–3 and close to an order of 1 for the most
concentrated formate solution (5 kmol m–3).
Table 2. Range of conditions for kinetic study in the batch reactor
(other conditions: cyclohexane as organic phase solvent; stirring
speed: 1100 rpm)
–3
Concentration of catalyst [kmol m
Temperature [K]
]
0.00132 to 0.005
303–313–323–333
0.024 to 0.144
Aq
Initial concentration of DMI
–3
in the organic phase [kmol m
Initial concentration of DMI
]
Org
0.0021 to 0.013
–3
in the aqueous phase [kmol m
]
Aq
Aqueous phase volume [10–6 m3]
20 to 100
10 to 20
0.2 to 1
0.2 to 5
Organic phase volume [10–6 m3]
Volume ratio α [m3Org
m ]
Aq
–3
Concentration of sodium formate
–3
in the aqueous phase [kmol m
TPPTS/Rh ratio
]
Aq
3.1 to 11.3
Figure 2. Influence of the initial substrate concentration on the ob-
served initial reaction rate riOrg for different sodium formate con-
Rh
–3
centrations (C
ϭ 0.002 kmol m , α ϭ 1, TPPTS/Rh ϭ 6,
Aq
Formate decomposition in molecular hydrogen and so-
dium hydrogen carbonate has been noted, with or without
substrate. This hydrogen production appears to be inde-
pendent of the catalytic cycle of the DMI reduction. Tests
have been performed under hydrogen pressures of 0.1 and
0.2 MPa to check for this, and no acceleration effect was
Aq
313 K); the points represent the experimental data and the curves
represent the semi-empirical model described by Equation 5
The effect of the phosphane on the initial rate exhibits a
noticed on the reduction rate. Thus, the DMI is not reduced negative order with respect to the phosphane concentration,
by the molecular hydrogen formed from formate decom- since the initial reaction rate decreases with increasing pho-
position.
sphane amounts in the aqueous phase (Figure 3).
1496
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502

Study of the H-Transfer Reduction of Dimethyl Itaconate by a Rh/TPPTS Catalyst
FULL PAPER
with α ϭ volumetric ratio of the organic phase and the
–3
aqueous phase (m3
m ).
Org Aq
The kinetic parameters of the above model were estim-
ated by means of a dynamic estimation and simulation soft-
ware, able to perform numerical integrations using the en-
tire concentration vs. time profiles and not only the initial
rates of reaction.
As shown in Figure 2 to Figure 5 and Figure 6, the model
fits rather well the experimental data except for low formate
Figure 3. Influence of the phosphane concentration on the ob-
DMI
–3
Rh
served initial reaction rate riOrg (C
ϭ 0.1 kmol m , C
ϭ
–3
Org
Org
Aq
concentrations (Ͻ 1 kmol m ). The results of the estima-
Aq
–3
HCOONa
–3
0.002 kmol m , α ϭ 1, C
ϭ 5 kmol m ); the points rep-
Aq
Aq
Aq
tion are presented in Table 3.
resent the experimental data and the curves represent the semi-
empirical model described by Equation 5
The influence of the sodium formate concentration was
also studied, and it seems more complex. As shown in Fig-
ure 4, a linear increase of initial rate is observed until the
–3
formate concentration reaches 1 kmol m . Above this con-
Aq
centration, the rate decreases with increasing formate con-
centration. Such a complex behaviour has already been ob-
[13]
´
served by Joo et al.
in the reduction of aldehydes by
water-soluble phosphane complexes of ruthenium in bi-
´
phasic medium. Joo evoked a salting-out effect due to the
high formate concentration to explain this result. Formate
concentration does indeed have a strong influence on the
partition coefficients of the substrate and product (see
Table 1).
Figure 5. Concentration vs. time profiles. Comparison between ex-
perimental data (symbols) and semi-empirical model (lines). (∆):
DMI
HCOONa
Standard test: T ϭ 313 K; C
ϭ 0.1; C
ϭ 5, TPPTS/
Org
Aq
Rh
Aq
Rh ϭ 6.3, α ϭ 1, C ϭ 0.002. Other tests: same with: ( ᭺) α ϭ
0.2; (᭜): α ϭ 0.53, CRAqh ϭ 0.005; (ᮀ): C
(0) ϭ 0.155, C
ϭ
DMI
HCOONa
Org
Aq
DMI
HCOONa
1; (᭿): C
(0) ϭ 0.05, C
ϭ 2.5; (᭹): TPPTS/Rh ϭ 3.1.
Org
DMI
Aq
HCOONa
–3
Rh
–3
Units: C
in kmol m ; C
and C in kmol m ; α in m
Org
Org
Aq
Aq Aq
3
–3
m
Org Aq
.
Figure 4. Influence of the sodium formate concentration on the
DMI
–3
Rh
observed initial reaction rate riOrg (C
ϭ 0.1 kmol m , C
ϭ
Org
Org
Aq
–3
0.002 kmol m , α ϭ 1, TPPTS/Rh ϭ 6); the points represent the
experimental data and the curves represent the semi-empirical
model described by Equation 5
Aq
All these new results led us to propose a new semi-empir-
ical model (Equation 4 and Equation 5), which represents
all the experimental results rather well.
DMI
DMI
Knowing that nDMI ϭ n
ϩ n
, and from the mass
Aq
Org
balance in the batch reactor we can obtain
Figure 6. Calculated vs. experimental initial rates of reaction; con-
HCOONa
ductions see Table 2 except for symbol (᭺) for which C
ϭ
Aq
–3
0.5 kmol m
Aq
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502
1497

N. Tanchoux, C. de Bellefon
FULL PAPER
Table 3. Estimated kinetic constants and statistical analysis for the ure 7c), showing no incorporation at this position. The two
diastereotopic protons H3 and H3Ј appear as two broad
kinetic model determined from 30 experiments
multiplets at δ ϭ 2.33 and 2.65, corresponding to an AB
Value[a]
Confidence interval[b]
system. Theses signals allowed us to determine a deuterium
incorporation of 58 and 47% for each proton, respectively,
i.e. a global quantitative deuterium incorporation at C3.
The methyl H1 appears as a broad multiplet at δ ϭ 1.14,
showing a 31% deuterium incorporation at C1. The
13C{1H}-NMR spectra and DEPT sequence of the product
confirmed there was no incorporation at the C2 position.
The triplet at δ ϭ 37 corresponds to a quantitative mono-
deuteration –CHD in the C3 position, confirmed by the
DEPT sequence. The multiplet at δ ϭ 16.7 shows an incorp-
oration at the C1 position. In the case of reduction with
H2O/DCOO–, the 1H-NMR spectra of DMS showed a
broad multiplet at δ ϭ 2.93 (Figure 7b) corresponding to
the proton H2, and showing no incorporation at this posi-
tion. The two proton H3 and H3Ј appeared as two doublets
of doublets at δ ϭ 2.4 and 2.72, corresponding to the coup-
ling with the methine proton. These signals showed no deu-
terium incorporation at this position. The methyl appeared
as a doublet and a small broad multiplet at δ ϭ 1.21, show-
ing 13% deuterium incorporation at C1. The 13C{1H}-
NMR spectra and DEPT sequence confirmed these results.
At δ ϭ 35.7 and 37.4, two singlets showed no deuterium
incorporation at C2 and C3, respectively. At δ ϭ 16.7, C1
appeared as a multiplet due to coupling with deuterium in-
corporated at this position.
k [m3Aq kmol–1 s–1]
0.053
2.1
0.048–0.058
1.8–2.4
K [m kmol–1]
–3
Aq
[a]
WRSS: weighted residual sum of squares for the parameter es-
[b]
timation: 1900. – Interval at 95% confidence level.
Labelling Studies
This study was completed by labelling experiments, in or-
der to determine the chemistry occurring, and maybe ob-
tain a better understanding of the mechanism of this reac-
tion. Table 4 gives the regioselectivity of the deuterium dis-
tribution in the products obtained by the reduction of
DMI, carried out with D2O/HCOO– first, then with H2O/
DCOO–. Deuterium was never incorporated in the methyl
ester groups.
Table 4. Amount of deuterium incorporated by the dimethyl
methylsuccinate
Attempted Complex Characterization in situ
The extent of deuterium incorporation was determined
by H-NMR spectroscopy using the signal of the methoxy
group as an internal standard (Figure 7).
The 31P-NMR spectrum of the catalytic complex in so-
dium formate shows broad signals centred at ca. δ ϭ 31
and –5, probably due to a fluxional process including free
1
TPPTS. Under catalytic conditions ([Rh] ϭ 0.013 kmol
–3
m
, TPPTS/Rh ϭ 4), i.e. in the presence of sodium form-
Aq
ate (3.3 kmol m–3) and DMI (0.02 kmol m–3), the NMR
spectrum shows two types of compounds. The singlet at δ ϭ
24.5 is assigned to a phosphonium according to previous
reports.[14,15] It is formed by nucleophilic attack of the free
TPPTS on the DMI, assisted by the Rhodium complex. In-
deed, as proved in a separate experiment, the formation of
the phosphonium is much slower when performed without
Rhodium. Thus a yield of ca. 10%, based on TPPTS, is
obtained without Rhodium, whereas a quantitative conver-
sion of the free TPPTS is evidenced in the presence of the
Rhodium complex. The change from fluxional to well de-
fined 31P spectra is thus attributed to the consumption of
free TPPTS by DMI under catalytic conditions, which leads
to the observation of the complex [Rh(TPPTS)3X] (X ϭ Cl
or OH)[16] based on the doublets and triplets at δ ϭ 35.5
and 52.4 (1JRh,P1 ϭ 147 Hz, 1JRh,P2 ϭ 182 Hz, and 2JP1,P2 ϭ
40 Hz), respectively. In water alone, under non-catalytic
conditions, signals due to the complex [Rh(TPPTS)3(OH)],
1
Figure 7. H-NMR spectra of DMS (a) and of the products from
reduction in DCOONa/H2O (b) and in HCOONa/D2O (c); signals
of the vinylic proton in the by-product DMM: (*) 5, (**) 6 (see
Table 5)
the phosphonium, and the TPPTS oxide are observed. A
1
In the case of reduction with D2O/HCOO–, the methine further signal at δ ϭ 29.8 (d, JRh,P ϭ 145 Hz) is assigned
proton H2 appears at δ ϭ 2.84 as a broad multiplet (Fig- to the cationic complex [Rh(TPPTS)2(cod)]ϩ.[17]
1498
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502

Study of the H-Transfer Reduction of Dimethyl Itaconate by a Rh/TPPTS Catalyst
FULL PAPER
allows the C2 position to remain untouched, without any
incorporation of deuterium, and only partial incorporation
in the methyl group when reduction is carried out with
H2O/DCOO–.
Discussion
Mechanism
The results of labelling studies differ considerably from
those obtained for mechanistic studies of slightly different
systems, i.e. formic acid as the hydrogen source, itaconic
acid as the substrate, or monophasic systems.[1–4]
The most surprising result was the absence of deuterium
incorporation in the C2 position while reducing the DMI in
either a D2O/HCOO– medium or in H2O/DCOO–. Very low
incorporation of deuterium in the C2 position has been re-
ported recently in the H-transfer reduction of dimethyl ita-
conate with a Rh/diphosphane catalyst in DMSO.[18] This
result indicates that the hydrogen atom incorporated in the
C2 position must come from the substrate dimethyl itacon-
ate itself.
The first step of the mechanism is a de-coordination of
one phosphane ligand to give a Rh complex containing two
phosphane equivalents. Step b is the coordination of the
olefin to the Rh complex. Step c is an intramolecular oxid-
ative addition of the olefin to the rhodium complex, which
gives a π-allyl rhodium hydride. This type of intermediate
is very well-known in the π-allyl mechanism of olefin iso-
merization with complexes of palladium[19] and rhodium.[20]
The formation of the phosphonium salt may well occur by
nucleophilic attack of free TPPTS at the coordinated π-al-
lyl, which supports the view of complex B being part of
the cycle.
The next step is an intramolecular rearrangement of a π-
allyl rhodium hydride complex to a rhodametallacyclobut-
ane. The formation of such a complex with addition of hy-
dride reagents to a π-allyl rhodium complex has already
been reported.[21,22] Steps b to d consist of the shift of a
hydrogen atom from position 3 to position 2. The resulting
sp2 carbon (C2) is reduced to an sp3 carbon intramolecularly
before any possible exchange with the reaction medium.
This mechanism is the only one we could propose that ex-
plains the results of the labelling experiments.
The mechanism of reduction is not likely to be close to
the one proposed by Sinou[2] for the hydrogenation of DMI,
as it was demonstrated that water did not play any role as
a reactant, which is not the case here. Indeed, after reduc-
tion with D2O/HCOO–, the product DMS presented a 31%
deuterium incorporation in the C1 position and a complete
monodeuteration in the C3 position. As it is known that the
exchange of hydrogen by deuterium in the formyl position
is very slow compared to hydrogen transfer,[1] the incorp-
oration of deuterium by the product cannot be attributed
to DCOO–, which is unlikely to be present in the catalytic
phase.
The formate then coordinates to the metallacyclobutane
intermediate. Step f of hydride transfer from formate to
When the reduction is carried out with H2O/DCOO–, a rhodium has already been postulated for the mechanism of
primary kinetic isotope effect of kH/kD ϭ 2.8 is measured. itaconic acid reduction by hydrogen transfer,[1] although it
This value, in agreement with those obtained by Brunner,[1] was not possible to decide whether it occurs externally or
clearly shows that the sodium formate is involved in the through an intramolecular pathway. It has been shown that
rate-determining step.
in this type of rhodium hydride or dihydride intermediate,
The only mechanism accounting for all these experi- scrambling always occurs.[1,23] That explains very well the
mental results seems to involve going through a rhodamet- partial incorporation of deuterium (13%) at the C1 position
allocyclobutane (Scheme 1). Only this type of intermediate when the reduction is carried out with H2O/DCOO–. The
Scheme 1
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502
1499

N. Tanchoux, C. de Bellefon
FULL PAPER
exchange Rh-D/H2O must be quite slow, as the incorpora- DMS using the Bodenstein approach for catalytic interme-
tion obtained in the C1 position (31%) is close to the theor- diates (Equation 6).
etical incorporation (33%) with D2O/HCOO–.
Step g is an intramolecular reductive elimination that op-
ens the metallacyclobutane and rapidly gives a rhodium al-
kyl intermediate. This type of equilibrium between a metal-
lacycle and a metal alkyl intermediate has already been seen
in such reactions as C–H activation,[24–26] the metal alkyl
intermediate being thermodynamically more stable than the
metallacycle hydride.[26]
Scheme 2
The final product is then recovered by hydrolysis, which
explains the quantitative incorporation observed at the C3
position when the reduction is carried out with D2O/
HCOO–. This result is confirmed by the lack of incorpora-
tion at C3 when H2O/DCOO– is used. Hydrolysis of alkyl
complexes of transition metals such as palladium (II)[27]
and selective incorporation of Hϩ in Rh-alkyl fragments[28]
has already been reported.
This rate law is in accordance with experimental results,
i.e. a negative order with respect to phosphane, first order
with respect to the catalyst concentration, and complex or-
der with respect to the substrate. It was not possible, how-
ever, to take into account the influence of the sodium form-
ate concentration. The semi-empirical model (Equation 4
and Equation 5) presents a negative order with respect to
the formate concentration, whereas the formal kinetic rate
law presents a positive order of between 0 and 1.
Although this expression gives a complex order with re-
spect to formate concentration, it was not possible to take
its influence into account over the whole range of concen-
tration used for the kinetic study.
The model is in good agreement with the experiments,
but it was not possible to estimate one set of kinetic para-
meters for both series of experiments with sodium formate
concentrations of 5 and 1 kmol m–3. In addition, as the
model possesses an intrinsic inaccuracy due to constants
coupling, it was not possible to determine values of intrinsic
kinetic constants with a sufficient accuracy, but only ratios
of constants. More tests under a wider range of conditions
where kinetic constants would not be coupled need to be
performed.
The by-product, dimethylmesaconate, is formed by an
isomerisation of the substrate DMS by migration of the
double bond from an external to an internal position. The
1
cis isomer, dimethylcitraconate, is not formed (13C and H
NMR). A similar secondary isomerisation reaction is re-
ported for the reduction of itaconic acid with formic acid/
triethylamine.[8] Curiously, in this case, the cis isomer, i.e.
citraconic acid, is formed. In the by-product, only the
methyl group (position 1) incorporates deuterium (10–15%
with H2O/DCOO– and 20–25% with D2O/HCOO–), which
is in accordance with the position of step i in the pro-
posed mechanism.
Kinetics
Choice of the rate determining step:
(i) The sodium formate is involved in or before the rate
determining step as a primary kinetic isotope effect is noted
when the reduction is carried out with H2O/DCOO–.
(ii) Hydrolysis occurs after the last rate determining step
as no isotope effect is measured when the reduction is car-
ried out with D2O/HCOO–.
(iii) Addition of sodium hydrogenocarbonate has no in-
fluence on the course of the reaction.[7] The CO2, a product
of the reaction, is not involved before the last rate determin-
ing step.
Conclusion and Future Work
(iv) The liberation of one equivalent of phosphane ligand
The kinetic and mechanistic study of H-transfer reduc-
occurs before the last rate determining step, as addition of tion of dimethyl itaconate by sodium formate/water under
phosphane inhibits the reaction. biphasic catalysis has been presented. A complex, semi-em-
The last rate determining step is then step f, as it occurs pirical kinetic model was first validated. It exhibits complex
after the formate coordination and before the liberation of order with respect to the substrate concentration, and in-
CO2 and the final, hydrolysis step. In addition, only the hibition by the phosphane. Numerical integration using the
steps involving terminal products have been considered. entire concentration vs. time profiles provided values for the
Thus, steps b to d involving intermediates A, B, and C have kinetic parameters of this model. Labelling studies and 1H-
been lumped in one step in the kinetics (Scheme 2), other- and 13C-NMR analysis were also performed. A catalytic
wise the kinetic rate law would have been more complex cycle has been proposed based on these results. A rhodame-
and would have induced more constants and larger uncer- tallacyclobutane intermediate best accounts for the system-
tainty in their estimation. The X letter used represents inter- atic non-deuteriation of the C2 position of the product, the
mediates A, B, and C (Scheme 2). The same process has reduction being carried out either by H2O/DCOO– or D2O/
been used for the steps e and f. The kinetic Scheme allowed HCOO–. The mechanism is also supported by the good fit
us to derive a formal kinetic rate law for the formation of between the empirical model and the formal rate law, de-
1500
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502

Study of the H-Transfer Reduction of Dimethyl Itaconate by a Rh/TPPTS Catalyst
FULL PAPER
data for the isotopomers identified in the organic layer: 1: δ ϭ 1.15
(d, 3J12 ϭ 7.1 Hz, 3 H, H1), 2.65 (dt, 3J32 ϭ 8.0 Hz, 2JHD ϭ 2.3 Hz,
1 H, H3), 2.80 (m, 1 H, H2), 3.61, 3.63 (s, 3 H, H5, H7). – 2: δ ϭ
rived from the mechanistic steps, despite a poor agreement
concerning the influence of sodium formate.
The main conclusion concerns the shift in the mechanism
on going from diacids to their ester derivatives. Unsaturated
acids reduction proceeds through a cis-1,2-hydrogen addi-
tion,[1,8] whereas this work shows that reduction of esters
should undergo a 1,3-addition.[1] This switch in mechanism
may well account for the lower optical yields observed in
the enantioselective H-transfer reduction of esters com-
pared to their corresponding acids.[8]
3
3
2
1.15 (d, J12 ϭ 7.1 Hz, 3 H, H1), 2.33 (m, J3Ј2 ϭ 6.0 Hz, JHD
ϭ
2.3 Hz, 1 H, H3Ј), 2.80 (m, 1 H, H2), 3.61, 3.63 (s, 3 H, H5, H7). –
3: δ ϭ 1.13 (dt, J12 ϭ 7.1 Hz, JHD ϭ 1.8 Hz, 2 H, H1), 2.65 (dt,
3
2
3J32 ϭ 8.0 Hz, JHD ϭ 2.3 Hz, 1 H, H3), 2.80 (m, 1 H, H2), 3.61,
2
3
2
3.63 (s, 3 H, H5, H7). – 4: δ ϭ 1.13 (dt, J12 ϭ 7.1 Hz, JHD
ϭ
1.8 Hz, 2 H, H1), 2.33 (m, J3Ј2 ϭ 6.0 Hz, JHD ϭ 2.3 Hz, 1 H,
H3Ј), 2.80 (m, 1 H, H2), 3.61, 3.63 (s, 3 H, H5, H7). – 5: δ ϭ 2.22
(d, 4J13 ϭ 1.5 Hz, 3 H, H1), 3.74, 3.70 (s, 3 H, H5, H7), 6.71 (broad,
3
2
1 H, H3). – 6: δ ϭ 2.22 (m, J13 ϭ 1.5 Hz, JHD ϭ 1.46 Hz, 2 H,
4
2
H1), 3.74, 3.70 (s, 3 H, H5, H7), 6.71 (broad, 1 H, H3).
Experimental Section
Labelling Study/Reduction of DMI in H2O/DCOONa: A solution
of [D2]formic acid in D2O (95% wt, 10 g, 0.19 mmol) was added to
an aqueous solution of sodium hydroxide (10, 20 cm3, 0.2 mmol).
The mixture was then evaporated and water was added to the res-
idue. This procedure was repeated twice to remove traces of D2O
and Dϩ. Water (40 cm3) was finally added to the thus prepared
[D]sodium formate. Preparation of the catalyst and realization of
General: 31P-NMR spectra (external reference H3PO4 85% D2O)
were recorded on a Bruker AM 300 (121.51 MHz) and a Bruker
1
AM 200 (81.015 MHz). H- and 13C-NMR spectra (external refer-
ence SiMe4) were recorded on a Bruker AM 200 (200.13 MHz and
50.32 MHz, for 1H and 13C, respectively). Deuterated water (99.9
atom% D, Aldrich) and [D2]formic acid (99ϩ atom%, 95% in D2O,
ACROS) for the labelling study were degassed and used without
further purification. – TPPTS 30% wt in water, sodium formate
(97%, Aldrich), dimethyl itaconate (97%, Aldrich), decane (99%,
Aldrich), mesaconic acid (99%, Aldrich), and the catalyst precursor
[Rh(cod)Cl]2 (98%, Strem) were used as received. All the experi-
ments were performed under argon or nitrogen. – Partition iso-
therms determination: The experimental procedure for determining
partition isotherms has already been described in a previous pa-
per.[7] – Kinetic studies in a well mixed batch reactor: All the details
have already been described in a previous paper.[7] The reduction
of dimethyl itaconate (DMI) was performed under a range of con-
ditions, given in Table 2. The initial rate of reaction is calculated
from the slope of concentration vs. time profiles at low conversion
(Ͻ 10%).
the kinetic tests were then performed as already described.[7]
–
After reduction the organic layer was separated, dried with CaCO3,
and the solvent was evaporated. The oily product was analyzed by
1
NMR spectroscopy. – 13C NMR: δ ϭ 16.7 (m, JCD ϭ 19.7 Hz,
1
C1), 35.7 (m, C2), 37.4 (s, JCD ϭ 20.0 Hz, C3), 51.7 (s, C7), 51.9
(s, C5), 172.3 (s, C6), 175.7 (s, C4).
The 1H-NMR data showed the presence of isotopomers 5, 6, 7,
3
2
and 8. – 7: δ ϭ 1.16 (d, J12 ϭ 7.1 Hz, 3 H, H1), 2.35 (dd, J33Ј
ϭ
16.3 Hz, J23Ј ϭ 6.0 Hz, 1 H, H3Ј), 2.67 (dd, J23 ϭ 8.0 Hz,1 H,
3
3
H3), 2.85 (m, 1 H, H2), 3.61, 3.63 (s, 3 H, H5, H7). – 8: δ ϭ 1.16
(dt, J12 ϭ 7.1 Hz, JHD ϭ 1.9 Hz, 2 H, H1), 2.35 (dd, J33Ј
ϭ
3
2
2
16.3 Hz, J23Ј ϭ 6.0 Hz, 1 H, H3Ј), 2.67 (dd, J23 ϭ 8.0 Hz, 1 H,
3
3
H3), 2.85 (m, 1 H, H2), 3.61, 3.63 (s, 3 H, H5, H7).
Synthesis of the By-Product DMM: A drop of sulfuric acid 98%
was added to a solution of mesaconic acid (2 g, 15.3 mmol) in
methanol (50 cm3). The mixture was stirred at room temperature
for 3 days. The solvent was then evaporated and the crude product
Labelling Study/Reduction of DMI in D2O/HCOONa: A solution
of TPPTS in water (30% wt, 2 cm3, 1.22 mmol) was evaporated.
D2O (2 cm3) was then added to the residue, and the procedure was
repeated twice. A solution of sodium formate in D2O (5, 8 cm3,
40 mmol) was then added to an orange slurry of [Rh(cod)Cl]2
(48.9 mg, 0.2 mmol) in the solution of TPPTS in D2O previously
prepared. The mixture was stirred at room temperature for about
12 h and stored at 4 °C. The catalytic run was performed as previ-
ously described.[7] After reduction, the organic layer was separated,
dried with CaCO3, and the solvent was evaporated. The oily prod-
uct was analysed by NMR spectroscopy. – 13C NMR: δ ϭ 16.7 (m,
1
obtained was characterised by H-NMR (200 MHz, CDCl3) spec-
1
4
troscopy. – H NMR: δ ϭ 2.30 (d, J13 ϭ 1.6 Hz, 3 H, H1), 3.78,
3.82 (s, 3 H, H5, H7), 6.80 (q, 1 H, H3).
Acknowledgments
JCD ϭ 19.7 Hz, C1), 35.5 (m, C2), 37.0 (t, JCD ϭ 20.0 Hz, C3), The authors are grateful to the Region Rhone-Alpes, the Centre
1
1
´
ˆ
51.5 (s, C7), 51.8 (s, C5), 172.1 (s, C6), 175.5 (s, C4). – 1H NMR National de la Recherche Scientifique and the Rhone-Poulenc com-
ˆ
Table 5. Different isotopomers/isomers present in organic layers after labelling experiments
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502
1501

N. Tanchoux, C. de Bellefon
FULL PAPER
[15]
[16]
[17]
[18]
M. Hernandez, P. Kalck, J. Mol. Catal. A: Chemical 1997, 116,
131–146.
W. A. Herrmann, J. A. Kulpe, W. Konkol, H. Bahrmann, J.
Organomet. Chem. 1990, 389, 85–101.
R. R. Schrock, J. A. Osborn, J. Am. Chem. Soc. 1971, 93,
2397–2407.
S. Lange, W. Leitner, R. Mynott, C. Wirtz, XIth International
Symposium on Homogeneous Catalysis, St. Andrews, Scot-
land, July 1998, p. 73.
G. W. Parshall, S. D. Ittel, Homogeneous Catalysis: The Ap-
plications and Chemistry of Catalysis by Soluble Transition
Metal Complexes, 2nd ed., Wiley-Interscience publication,
1993, p. 14–16.
K. E. Koenig, W. S. Knowles, J. Am. Chem. Soc. 1978, 100,
7561–7564.
´
pany for financial support. N. T. is also grateful to the Region
ˆ
Rhone-Alpes for a PhD grant.
[1]
W. Leitner, J. Brown, H. Brunner, J. Am. Chem. Soc. 1993, 115,
152–159 and references therein.
[2]
A. Bucsai, J. Bakos, M. Laghmari, D. Sinou, J. Mol. Catal. A:
Chemical 1997, 116, 335–342.
[3]
D. Heller, R. Kadyrov, M. Michalik, T. Freier, U. Schmidt, H.
[19]
W. Krause, Tetrahedron: Asymmetry 1996, 7, 3025–3035.
[4]
A. Harthun, R. Kadyrov, R. Selke, J. Bargon, Angew. Chem.
Int. Ed. Engl. 1997, 36, 1103–1105.
[5]
[20]
[21]
[22]
[23]
[24]
[25]
K. T. Wan, E. Davies, Tetrahedron: Asymmetry 1993, 4, 2461–
2468.
[6]
R. A. Periana, R. G. Bergman, J. Am. Chem. Soc. 1984, 106,
7273–7274.
R. A. Periana, R. G. Bergman, J. Am. Chem. Soc. 1986, 108,
7346–7355.
Y. Amrani, L. Lecomte, D. Sinou, Organometallics 1989, 8,
542–547.
[7]
C. de Bellefon, N. Tanchoux, S. Caravieilhes, D. H. Schweich,
Catal. Today 1999, 48, 211–219.
[8]
D. J. Hardick, I. S. Blagbrough, B. V. L. Potter, J. Am. Chem.
Soc. 1996, 118, 5897–5903.
H. Brunner, W. Leitner, J. Organomet. Chem. 1990, 387, 209–
217.
T. H. Tulip, D. L. Thorn, J. Am. Chem. Soc. 1981, 103, 2448–
[9]
X. Wang, L. K. Woo, J. Mol. Catal. A: Chemical 1998, 130,
2450.
171–176.
W. D. Jones, F. J. Feher, J. Am. Chem. Soc. 1984, 106, 1650–
1663.
[10]
T. Karlen, A. Ludi, Helv. Chim. Acta 1992, 75, 1604–1606.
C. de Bellefon, N. Tanchoux, S. Caravieilhes, J. Organomet.
[11]
[26]
[27]
R. H. Crabtree, Chem. Rev. 1985, 85, 245–269.
S. Ganguly, D. Roundhill, Organometallics 1993, 12, 4825–
4832.
Chem. 1998, 567, 143–150.
[12]
R. A. W. Johnstone, A. H. Wilby, Chem. Rev. 1985, 85, 129–
170.
[28]
C. Detellier, G. Gelbard, H. B. Kagan, J. Am. Chem. Soc. 1978,
[13]
´
´
A. Benyei, F. Joo, J. Mol. Catal. 1990, 58, 151–163.
100, 7556–7561.
[14]
C. de Bellefon, E. Pollet, P. Grenouillet, J. Mol. Catal. A:
Received August 27, 1999
Chemical 1999, 145, 121–126.
[I99314]
1502
Eur. J. Inorg. Chem. 2000, 1495Ϫ1502